Proof of physical work

Where the term comes from

Tushar Jain and Shayon Sengupta of Multicoin Capital named the thesis in April 2022. Their argument was that crypto-economic protocols could incentivise people anywhere to do verifiable work that builds real-world infrastructure, and that this changes how such infrastructure gets financed: labour cost in the Helium network is $0 to the protocol, because the network is built by independent hosts rather than employees.¹

Their two examples set the range. Helium had passed 600,000 hotspots thirty months after launching in August 2019, against 417,000 cell towers across the entire US telecommunications industry. Hivemapper outsourced street-level mapping to drivers who already owned adequate hardware, rather than running a fleet of specialist vehicles, and the same essay noted the approach extends to air pollution, wireless coverage quality, noise and weather data.¹

What has to be proved

A 2026 review of DePIN tokenomics in Frontiers in Blockchain puts verification of physical work first among the four recurring design primitives, defining it as a reliable, automated protocol for proving that contributors are actually providing the physical service they claim. It cites Filecoin's proof of replication and proof of spacetime for storage and Helium's proof of coverage for wireless as the worked instances.²

The same review pairs verification with the mechanism that gives it teeth: providers stake the native token as collateral, and that stake can be partially or fully slashed for poor performance or dishonest reporting.² Verification tells you the work did not happen. Collateral is what makes the finding cost something. A design with the first and not the second has a detector and no consequence.

Proving presence versus proving output

The most instructive development in this category is that the network which popularised proof of physical work has retired its own version of it. HIP-149, approved in 2026, removes proof of coverage from both the Mobile and IoT verifier oracles and pays Mobile data deployers pro rata of rewardable bytes instead.³

The proposal's reasoning is worth quoting closely: proof of coverage pays for existing rather than for serving subscribers, and the work that grows the network is offload utility rather than coverage proofs.³ That is a network stating, seven years in, that it had been paying for the wrong thing. Proving a radio is where it says it is turns out to be a proxy for value, and the moment real traffic exists, the traffic itself is a better proof than any challenge protocol.

The design consequence is simple to state and unpopular to adopt. If your service produces a measurable output that customers pay for, reward the output. Presence proofs are for the period before there is any output to measure, and they should be scheduled to end.

Where these designs get gamed

Every proof of physical work scheme is an incentive to produce whatever the proof measures at the lowest possible cost. If the proof is a radio signal witnessed by neighbours, colocated devices and spoofed signals become the cheapest strategy. If it is imagery, the cheapest strategy is repeated capture of the same easy road. If it is storage, it is deduplication against a challenge the protocol reuses.

Assume adversarial behaviour from the first epoch, not after the first incident. Score work against demand rather than against effort, so the cheapest gaming strategy is also the least rewarded. Vary challenges so they cannot be precomputed. Require redundant, independent evidence for anything expensive. And keep stake at risk proportional to the reward a dishonest operator could earn before detection, because a slashing rule smaller than the exploit is a fee, not a deterrent.